OpenAI has announced that it is experimenting with a feature that allows ChatGPT to remember both information about individual users and how they want the chatbot to respond to different types of queries.

When chatting with the bot, users can ask it to remember specific things like we often ask Google to do or allow ChatGPT to pick up details itself and these details can be used to inform how ChatGPT responds to you, be it in the form of ideas and suggestions it generates to the formats it delivers answers in.

OpenAI has said in a blog: “If you want ChatGPT to forget something, just tell it. You can also view and delete specific memories or clear all memories in settings (Settings > Personalization > Manage Memory). ChatGPT’s memories evolve with your interactions and aren’t linked to specific conversations. Deleting a chat doesn’t erase its memories; you must delete the memory itself.”

The feature is rolling out to a small portion of ChatGPT free and Plus users this week and a broader roll out will take place soon.

At the moment, large language models usually use two types of memory: one baked into the AI model during the training process and an in-context memory (conversation history) that remains during a session. Once a new session begins, ChatGPT forgets what you have told it during a conversation. The company hasn’t explained the technique that it is going to use to retain “memory”.

There are several applications for the memory feature, like having one’s preference for how meeting notes are required to be formatted and retaining information about a subject you often talk about. The new feature can be helpful for businesses. ChatGPT can be used to remember your tone and voice. Memories in the ChatGPT “workspace” are excluded from training AI models, the company says. Another area where this may help is recommending your next book. With GPTs that come with a distinct memory, Book GPT (which recommends your next book) can remember your favourite genres, and will offer recommendations based on that information.

The memory feature can be turned off and the company said: “We’re taking steps to assess and mitigate biases, and steer ChatGPT away from proactively remembering sensitive information, like your health details — unless you explicitly ask it to.”

The new feature appears to be an expansion of the ability to add instructions and personal preferences, which the company allowed last year. ChatGPT can now draw on a much wider and more detailed array of information.

Memory will be an important feature for ChatGPT and it’s a bit like how Internet services learn about us — what we click or like, helping develop a profile of us over time. By default, memory will be turned on, and OpenAI says memories could be used to train its models going forward. But companies using ChatGPT Enterprise and Teams won’t have their data sent back to the models.

Meanwhile, the Gemini app, formerly known as Bard, promises to make interacting with your smartphones and other devices considerably smarter and easier. Google has updated its privacy policy for Gemini and details how long the company will keep conversations that are “reviewed or annotated” by human reviewers: “Conversations that have been reviewed or annotated by human reviewers (and related data like your language, device type, location info, or feedback) are not deleted when you delete your Gemini Apps activity because they are kept separately and are not connected to your Google Account. Instead, they are retained for up to three years.” ChatGPT lets you permanently delete conversations every 30 days.